You realize LLMs are not that smart when you ask for a time estimate for a task and their knowledge being on what humans can do they grossly overestimate how long it's gonna take. Sometime they tell you 2 days and they can do it under 2h

V4 flash as the default agent / interface is very good and very fast though.
Then use other models when you need raw powers.
I started using Hermes Agent today, too soon to tell but pretty good so far.

First trade idea that came out of this is long JPY, with a 3 months ish horizon, I'm gonna buy a few calls to make it symbolic but even if they become multi baggers it won't move the needle for my pf

NQ / XYZ perp contract, this is the new HLP / default yield on HL

Interesting timing for this headline with VIX at 14 and every tech stock up 20%+ from the lows

And for coding tasks that are not in the top 5% of difficulty you can ask V4 0731 to create the code cheaply, have SOL 5.6/Kimi/ Fable review it and have again V4 implement modifs + last review from SOTA model.
You use maybe 75% less token that way for result a that is 99% the same

Very easy to do with Hermes where your subscription can easily pull credits from any model

Ceasefire but everytime we locate and can kill an ennemy we do it anyway

ISRAELI MILITARY SAYS IT KILLED A PROMINENT BADR UNIT COMMANDER AFFILIATED WITH HEZBOLLAH IN DEIR AL-ZAHRANI. ...

In 1h and at a cost of 10$ a month I built an algo to find me the cheapest tickets to anywhere on earth in one querry, maybe 2. Incredible stuff

Gone are the days working against google flights / kayak

Most companies will either become data pipes or go bankrupt

Your UX is nothing agains the power of even current day LLM

I longed some YEN, will add if we go lower

My conviction increased in that trade so I will make it a 50% normal sized position

(If I am given a better a better entry)